DEPR: Styler.where in favour of using applymap and apply · Issue #40124 · pandas-dev/pandas (original) (raw)

Styler by definition is a conditional renderer. The functions it provides, apply and applymap, serve as conditional functions to render a style.

The function Styler.where therefore, in the first case, seems unnecessary and indirectly confusing since it is a conditional rendering function implying perhaps the other methods are not. Not only is it just a 'convenience' wrapper of applymap:

def where(cond, value, other, subset, **kwargs):
    return self.applymap(lambda v: value if cond(v) else other, subset=subset, **kwargs)

But I also think, in the second case, it is quite inconvenient for a couple reasons:

1. The docs are ambiguous:

cond: callable: cond should take a scalar and return a boolean.
**kwargs: dict: pass along to cond.

and in fact kwargs appear in the code but do nothing.

1. It feels that there is more benefit to be had by a user learning how you use apply and applymap to the same effect through good examples.

For example:
styler.applymap(lambda v: props if cond(v) else None, subset=subset) or
styler.apply(lambda d: np.where(d > 10, props, ""), axis=None, subset=subset)

seem pretty straightforward replacements???

Perhaps documenting in the first instance that apply and applymap are preferred over where is non-controversial.